Dr. Mehmet Turan is instructor for the following graduate courses at the Bogazici University:


Course Description: While deep learning has achieved remarkable success in supervised and reinforcement learning problems, such as image classification, speech recognition, and games, these models are, to a large extent, specialized to the single task for which they are trained. This course addresses the setting in which multiple tasks are to be solved, and explores how the structure that results from multiple tasks can be used to learn more efficiently or effectively. These include goal-based reinforcement learning techniques that use the structure of the provided goal space to learn many tasks much faster, meta-learning methods that aim to learn efficient algorithms that can learn new tasks quickly, curriculum and lifelong learning, where the problem requires learning a sequence of tasks using their common structure to enable knowledge transfer. 


Course Description: This lecture is about fundamentals of deep learning and its applications in medical image analysis. Deep learning studies data-driven model generation to make predictions about data. While traditional methods successfully solve problems where final value is a simple function of input data, deep learning techniques are able to capture composite relations between non-immediately related fields. Recent developments in deep neural networks have catalyzed research in many fields, including medical image analysis. This course aims to cover the fundamentals of deep learning and survey the contemporary research and state-of-the-art architectures in medical anomaly detection, segmentation, classification and grading.